neilsharma425

5 ms·

1.
Show HN: EvalsHub: Your AI is failing in production and you don't know it(evalshub.ai)
4 points by neilsharma425 7mo ago|1 comments
2.
by neilsharma425 7mo ago
Has anyone found a working workaround yet? I use dnsmasq for .local dev routing and held off updating after seeing this but curious if there is a viable path forward short of waiting for Apple to patch it.
3.
by neilsharma425 7mo ago
The phone broker model exists largely because the plan comparison UI is genuinely unusable for most people, so there is a real problem here worth solving. Curious how you handle the subsidy estimation step. That calculation has a lot of edg
4.
by neilsharma425 7mo ago
Cute idea. The constraint that all letters must be used is what makes this interesting over just a free-form crossword builder since it turns it into a proper puzzle. How do you handle validation? Specifically, are you checking connectivity
5.
by neilsharma425 7mo ago
Neat problem to work on. The tail number lookup is the hard part and it sounds like you solved it the right way, by finding the people who actually track this obsessively rather than trying to scrape it yourself. Two questions: how stale do
6.
by neilsharma425 7mo ago
10:30 AM The Termite bundling is the most interesting part. Packaging embedding and reranking inference alongside the database means no separate model server to manage and no network hop for every vector op. Curious about resource contentio
7.
by neilsharma425 7mo ago
The "specific grievance" detail is what makes this interesting. Most multi-agent sims feel flat because agents are just goal-oriented — giving them a pre-existing tension before the scene starts is a much more realistic model of h